Classifications

• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F9/00—Arrangements for program control, e.g. control units
  • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
  • G06F9/46—Multiprogramming arrangements
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F9/00—Arrangements for program control, e.g. control units
  • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
  • G06F9/46—Multiprogramming arrangements
  • G06F9/52—Program synchronisation; Mutual exclusion, e.g. by means of semaphores

Definitions

• Computer systems are increasingly networked with one another via communication devices. In some applications, particular attention must be paid to the synchronization of the data processing systems combined in a network.
• the synchronicity required by the computers of a network can be predominantly of a temporal nature.
• the computers should work as parallel as possible, i.e. are always in the same processing state as far as possible with respect to the processing of the commands of a current program.
• the computers should therefore process the same processing sequence whenever possible.
• the synchronicity of computers in a network can result in the additional requirement that the respective processing sequences in the computers should also have the same data-technical meaning as possible.
• the computers of a network should achieve the same possible result values when the command processing takes place at approximately the same time on the basis of as identical as possible initial values.
• the comparison of the initial values or selected current processing values of the individual programs on the computers is also advantageous in the context of the temporal synchronization.
• two connected computers with identical programs should have the same output data, e.g. Process measured values under the boundary condition of high computer system availability. It must be regularly ensured that the processing status of the two computers may be due to Even slightly different processing speeds do not diverge too much in the medium term. Furthermore, it must be regularly ensured that both computers have matching processing results. This is a prerequisite for the fact that in the case where the two computers e.g. be used to control a technical process that requires high availability and one of the two computers fails, the other computer can continue to control the technical process virtually without jump. In such a "one of two" system, the two computers involved must be synchronized in time by special synchronization measures. Furthermore, their current processing contents must be checked regularly for equality.
• Comparable boundary conditions are also present in a so-called “two out of three” system.
• this system the requirements of high availability of the system on the one hand and high processing security of the data on the other hand can be fulfilled at the same time.
• three connected computers can be used The same output data must be processed with the aid of identical programs, in which case it must also be ensured at regular intervals that the processing states of the three computers do not diverge too much and that, in particular, the same processing results are available in each computer If processing results detect inequalities, one computer can be regarded as faulty and must leave the network if its current processing result suddenly and permanently deviates from the matching processing results of the other two computers As if one of the three computers failed, both processing reliability and availability.
• the invention is therefore based on the object of specifying a method for synchronizing at least two programs, in particular user programs, on computers connected to one another which is as simple, robust and, in particular, requires little processing time.
• FIG. 3 shows a first time schedule, on which the sequence of the synchronization method according to the invention is explained using the example of two computers in a network which can form a so-called "one of two” or "two of two” computer system, and
• FIG. 1 shows a first example of a communication link between two data processing systems, hereinafter abbreviated to computers.
• a first computer A which can be operated via a keyboard and a screen 4, is connected via a first coupling 3, and a second computer B, which can be operated via a keyboard and a screen 6, via a second coupling 6 with a data bus 1 connected.
• a standing communication link 8 is set up between these two computers, which is indicated in FIG. 1 by a dotted line 8. Via such a fixed, connection-oriented communication link, data can be continuously exchanged directly between the two computers.
• Each computer is set up to send and receive data exactly from the other computer.
• FIG. 2 shows a further example of a communication connection between the two data processing systems A and B.
• No communication connection is permanently established between the two computers. Rather, this communication is based on the so-called "broadcast principle".
• computer A feeds a message 10, ie a so-called broadcast, into data bus 1 when required.
• All other computers connected to data bus 1, in the example of the figure 1, the computer B and possibly other computers (not shown in FIG. 1) are in an operating mode which enables data to be received via the data bus 1. All other computers initially receive this Broadcast fed into the data bus by computer A and then check whether the data of the broadcast are addressed to them or have usable content.
• the synchronization method according to the invention can be used without restriction at least for the two types of connection shown in FIGS. 1 and 2 and will be explained in detail below using the example of FIG. 3.
• a general time axis 14 with time stamps Tl ... TU is shown on the left side.
• These points in time represent the starting points of processing states of the two computers A and B, which occur when the synchronization method according to the invention is carried out.
• These processing states are represented for computer A on a time axis 15 to the right of the general time axis 14 and for computer B on a time axis 15 lying to the far right of FIG. 3 in the form of function boxes.
• the time axes 15, 16 for computers A, B symbolize the sequence or progress in the
• the processing states required for the synchronization method according to the invention are inserted into the natural sequence of processing the commands from task A and B and thus each cause an at least temporary interruption in the processing sequence of the command instructions from task A and B.
• the method according to the invention has the advantage that the additional processing states caused by the synchronization and inserted into the course of the individual tasks are as short as possible.
• the two tasks running on the computers A, B are therefore only interrupted for the purpose of mutual synchronization for as long as is absolutely necessary.
• the synchronization process triggered at time T2 is therefore necessary. According to the invention, this is initiated so that the computer of the network, in the example of FIG. 3, computer A, is the first to reach a synchronization point during the processing of the corresponding task, that is to say, when processing the command, it reaches a command that synchronizes with another computer or the other computers in the network.
• the further processing of the commands of the current user program is first interrupted by computer A and at least one set of synchronization data for at least one other computer in the network is Interface fed. In the example in FIG. 15, this is represented by processing block 17, which is labeled "send syncdata x" and has an arrow 35 going away from it.
• the computer which triggers the synchronization switches immediately to a reception state immediately after the synchronization data set has been fed in, i.e. in a mode of operation that enables the receipt of corresponding data records from programs on other computers in the network.
• the program of the sending computer is not expected to receive acknowledgment messages from the programs on receiving computers of the network for completing the feeding-in process.
• a switch is immediately made to a data reception state, and the further processing of the current task is interrupted by a program of another computer in the network until a corresponding synchronization data set in particular arrives.
• this is symbolized by the processing block 19 labeled “receive wait” on the time axis 15 of computer A.
• the scope of the data to be transferred can be significantly reduced by means of a so-called signature formation.
• signature formation a data record is opened by a type of compression reduced a unique identifier, which can have a much smaller size compared to the data set.
• Computer B Symbolizes time axis 16 of computer B and an arrow 36 emanating therefrom.
• Computer B preferably switches to a data receiving state immediately after the synchronization data record has been fed in. Since the other computer A of the network in the example already fed in its synchronization data record at time T2, after switching from computer B to the processing state “receive”, no “wait” phase occurs, but the synchronization data record that is already pending can syncdata x can be received directly from computer A. This is symbolized in FIG. 3 at time T8 by the processing block 26 labeled “receive / receive syncdata x” on computer B's time axis 16.
• Time axis 15 by processing block 24 "processing sync data y", and computer B this is shown on time axis 16 by processing block 28 "processing sync data x". These two edits now also run almost simultaneously.
• the content of the respective synchronization data record plays no role in the synchronization method according to the invention.
• the mutually transmitted synchronization data sets syncdata x and syncdata y can each consist of a single bit, for example. This has the function of a time stamp.
• a temporal synchronization of the processing states of computers can already be achieved in this way using the method according to the invention.
• the synchronization data records can of course contain any additional information that can be used, for example, to compare the processing results of the two computers. In the example in FIG.
• processing syndata y the processing of the instruction instructions of the interrupted task can be continued A are triggered ("return to / ...: task A"), especially if a match between the two synchronization data records has been detected. On the other hand, the further processing interrupted or no longer recorded (“... / stop: task A").
• processing syndata y the processing of the instruction instructions of the interrupted task can be continued A are triggered ("return to / ...: task A"), especially if a match between the two synchronization data records has been detected.
• the further processing interrupted or no longer recorded (“... / stop: task A").
• the method according to the invention has the particular advantage that synchronization can take place particularly quickly, since none of the computers in the network which feed synchronization data into the data interface, e.g. the acknowledgment of the correct receipt of this data by another computer in the network is necessary and expected. Rather, another computer in the network virtually "answers", in which it itself feeds synchronization data into the data interface. In this way, there is minimal time loss and a minimal additional load on the data transmission link between computers in a network due to the additional exchange In the example in FIG.
• this synchronization that has been achieved or restored is achieved by a further synchronization, at the time TU almost simultaneously, both in the course of the instruction instructions from task A on the time axis 15 and in the sequence of the instruction instructions "B syncpoint x + 1" or "syncpoint y + 1" are displayed by task B on the time axis 16.
• an expiry time is advantageously started by a computer after at least one set of synchronization data has been fed in for at least one other computer in the network, and one computer in the event that up to At the end of the expiry time, no set of synchronization data could be received from at least one other computer, specified command instructions were carried out, in particular a message was generated or normal command processing was continued.
• This measure serves to ensure, on the other hand, that a computer requesting synchronization does not constantly and possibly in vain wait for the arrival of a synchronization data record from this other computer when an error occurs in another computer in the network. If the computer requesting synchronization by feeding in its own synchronization data record could not receive a synchronization data record from another computer when the specified expiry time had expired, suitable measures can be initiated by the requesting computer. For example, you can try the
• Synchronization with other computers in the network is interrupted and the processing of the commands of the current user program which is suspended for the duration of the synchronization attempt is continued.
• the sequence of the synchronization method according to the invention is explained.
• the synchronization method according to the invention can also be used without restriction for a larger number of computers interconnected in a network.
• the sequence of a user program “task A” and in particular a synchronization routine “sync routine A” constructed in accordance with the invention are in the left area in a first processing block for computer A, in the central area of FIG. 4 in one second processing block for computer B the execution of a user program “task B” and in particular a synchronization routine “sync routine B” constructed according to the invention and finally in the right area of FIG.
• a third processing block for computer C the execution of a user program “task C” "and in particular a synchronization routine” sync routine C "constructed according to the invention.
• the computers are coupled to one another in a network, in particular via a data network.
• the user programs task A, task B and task C can consist of identical instruction instructions, so that the network shown in FIG. 4 then represents a so-called "two out of three" computer network system.
• computers A, B and C each reach a program point in which synchronization with the other computers in the network is requested and triggered.
• these are the program point "syncpoint X" in the course of "task A”, the program point “syncpoint Y” in the course of "task B” and the program point “syncpoint Z" in the course of "task C”.
• Each of these program items is followed by a error routine with the designation "write syncdata x", "write syncdata y” or "write syncdata z".
• the synchronization routine "sync routine A" from computer A has the following command instructions in the example of FIG. 4:
• the synchronization routine “sync routine B” from computer B has the following command instructions in the example of FIG. 4:
• Computer A feeds a synchronization data record x, which is directed to the computer C of the network, into the data interface (see dotted arrow with reference number 29).
• Computer A feeds a synchronization data record x, which is directed to the computer B of the network, into the data interface (see solid arrow with reference number 32).
• Computer A starts an expiry time A within which the synchronization telegrams y, z of the other computers B, C of the network are expected to be received.
• Computer A switches to a receive state and interrupts further command processing.
• Computer A receives a synchronization data record addressed to it and fed in by computer B. y (see solid arrow with the reference symbol 31).
• Computer A remains in the receive state and continues to interrupt command processing.
• Computer A receives a synchronization data record z directed to it and fed in by computer C ⁇ see dash-dotted arrow with reference symbol 30).
• Computer A resets the expiry time A before the predetermined expiry time is reached, since in the meantime the expected synchronization data records from computers B and C could be received in time.
• Computer A carries out processing operations of its own synchronization data record x and the received synchronization data records y, z.
• the content of the synchronization data records x, y, z is checked for the presence of identity. If the desired result occurs during the processing operations, the further processing of task a is continued. Otherwise, command processing from task A to computer A remains interrupted and application-specific error handling can be triggered.
• Computer B feeds a synchronization data record y, which is directed to computer A of the network, into the data interface (see solid arrow with reference number 31).
• Computer B feeds a synchronization data record y, which is directed to the computer C of the network, into the data interface (see solid arrow with reference number 34). 12. Computer B starts an expiry time B within which the synchronization telegrams x, z of the other computers A, C of the network are expected to be received.
• Computer B switches to a receive state and interrupts further command processing.
• Computer B receives a synchronization data record z addressed to it and fed in by computer C (see solid arrow with reference numeral 33).
• Computer B remains in the receive state and continues to interrupt command processing.
• Computer B receives a synchronization data record x addressed to it and fed in by computer A (see solid arrow with reference numeral 32).
• Computer B resets the expiry time B before reaching the specified expiry time, since in the meantime the expected synchronization data records from computers A and C could be received in good time.
• Computer B carries out processing operations of its own synchronization data set y and the received synchronization data sets x, z. In particular, the content of the synchronization data records x, y, z is checked for the presence of identity. If the desired result occurs during the processing operations, the further processing of task b is continued. Otherwise the command processing from task B on computer B remains interrupted and it can be application-dependent
• Computer C feeds a synchronization data record z, which is directed to computer B of the network, into the data interface (see solid arrow with reference number 33).
• Computer C feeds a synchronization data record z, which is directed to computer A of the network, into the data interface (see dash-dotted arrow with reference numeral 30).
• Computer C starts an expiration time C within which reception of the synchronization telegrams x, y from the other computers A, B of the network is expected.
• Computer C switches to a receive state and interrupts further command processing.
• Computer C receives a synchronization data record x addressed to it and fed in by computer A (see dotted arrow with reference symbol 29).
• Computer C remains in the receiving state and continues to interrupt command processing.
• Computer C receives a message addressed to it from
• Computer B fed synchronization data set y (see dotted arrow with reference numeral 29).
• Computer C resets the expiry time C before reaching the predetermined expiry time, since in the meantime the expected synchronization data records from computers A and B could be received in good time.
• Computer C carries out processing operations of its own synchronization data record z and the received synchronization data records x, y.
• the content of the synchronization data records x, y, z is checked for the presence of identity. If the desired result occurs during the processing operations, the further processing of task c is continued. Otherwise the command processing from task C on computer C remains interrupted and application-specific error handling can be triggered.
• each computer A, B or C each sends one to the feeds another set of synchronization data to another computer in the network, and this computer expects to receive a corresponding set of synchronization data from the other computers in the network in an order reverse to the previous feed.
• Computer A each send a synchronization data record x directed to computer C and computer B (send C, x and send B, x).
• the computer A is then expected to receive the synchronization data sets y and z from the computers B and C (receive B, y and receive C, z).
• the synchronization method according to the invention can also be used unchanged if the tasks of the connected computers do not consist of identical sequences of command instructions, but are only set up to exchange data in certain program points or even only a computer on the Receiving data from the other computer is set up.

Landscapes

• Engineering & Computer Science (AREA)
• Software Systems (AREA)
• Theoretical Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Engineering & Computer Science (AREA)
• General Physics & Mathematics (AREA)
• Multi Processors (AREA)
• Hardware Redundancy (AREA)

Applications Claiming Priority (3)

Publications (2)

Family

ID=7795027

Family Applications (1)

Country Status (9)

Families Citing this family (18)

Family Cites Families (14)

• 1996
  • 1996-05-22 DE DE19620622A patent/DE19620622A1/de not_active Withdrawn
• 1997
  • 1997-05-21 AT AT97926963T patent/ATE186790T1/de not_active IP Right Cessation
  • 1997-05-21 DE DE59700733T patent/DE59700733D1/de not_active Expired - Lifetime
  • 1997-05-21 JP JP09541383A patent/JP2000510976A/ja active Pending
  • 1997-05-21 ES ES97926963T patent/ES2140241T3/es not_active Expired - Lifetime
  • 1997-05-21 WO PCT/DE1997/001023 patent/WO1997044731A1/fr not_active Application Discontinuation
  • 1997-05-21 EP EP97926963A patent/EP0898744B1/fr not_active Expired - Lifetime
  • 1997-05-21 KR KR1019980709369A patent/KR20000015819A/ko not_active Application Discontinuation
  • 1997-05-21 CN CN97194435A patent/CN1217798A/zh active Pending
• 1998
  • 1998-11-23 US US09/197,509 patent/US6308223B1/en not_active Expired - Lifetime

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events